Scale-aware monophonic retune over the framework's YIN detector and phase-vocoder shifter. More...

#include <PitchCorrector.h>

Public Member Functions
void	prepare (const AudioSpec &spec, int fftSize=0)
	Allocates the detector and shifter state (setup thread).

void	reset () noexcept
	Clears all signal state and keeps the parameters (stream owner).

void	setScale (std::uint16_t scaleBitmask, int rootPitchClass) noexcept
	Selects the scale the output snaps to.

void	setRetuneSpeedMs (T ms) noexcept
	Sets the retune speed: the time constant, in milliseconds, of the glide from the sung pitch to the target note.

void	setFormantPreserve (bool on) noexcept
	Enables the shifter's cepstral formant preservation (the anti-chipmunk envelope pre-warp). Default off.

std::uint16_t	getScaleMask () const noexcept

int	getRootPitchClass () const noexcept

T	getRetuneSpeedMs () const noexcept

bool	getFormantPreserve () const noexcept

std::vector< uint8_t >	getState () const
	Serializes the parameter state (setup/UI threads; allocates).

bool	setState (const uint8_t *data, size_t size)
	Restores parameters from a blob (tolerant; rejects foreign ids). A missing field restores its default: chromatic from C, hard snap, formants off.

int	getLatency () const noexcept
	Reports the signal latency in samples: twice the analysis frame, which is 4096 samples at 44.1/48 kHz and about 85 ms at every supported rate, since the frame follows the rate. Zero before prepare().

int	getFrameSize () const noexcept
	The analysis frame in effect, in samples: the automatic choice for this rate, or the rounded explicit request. Zero before prepare().

void	processBlock (AudioBufferView< T > buffer) noexcept
	Processes audio in-place. Pass-through until prepare() succeeds.

Detailed Description

template<FloatType T>
class dspark::PitchCorrector< T >

Scale-aware monophonic retune over the framework's YIN detector and phase-vocoder shifter.

Template Parameters

T	Sample type (float or double).

Definition at line 252 of file PitchCorrector.h.

Member Function Documentation

◆ getFormantPreserve()

template<FloatType T>

bool dspark::PitchCorrector< T >::getFormantPreserve ( ) const

inlinenoexcept

Returns: Whether formant preservation is requested.

Definition at line 397 of file PitchCorrector.h.

◆ getFrameSize()

template<FloatType T>

int dspark::PitchCorrector< T >::getFrameSize ( ) const

inlinenoexcept

The analysis frame in effect, in samples: the automatic choice for this rate, or the rounded explicit request. Zero before prepare().

Definition at line 437 of file PitchCorrector.h.

◆ getLatency()

template<FloatType T>

int dspark::PitchCorrector< T >::getLatency ( ) const

inlinenoexcept

Reports the signal latency in samples: twice the analysis frame, which is 4096 samples at 44.1/48 kHz and about 85 ms at every supported rate, since the frame follows the rate. Zero before prepare().

Definition at line 430 of file PitchCorrector.h.

◆ getRetuneSpeedMs()

template<FloatType T>

T dspark::PitchCorrector< T >::getRetuneSpeedMs ( ) const

inlinenoexcept

Returns: The retune speed in milliseconds (0 = hard snap).

Definition at line 391 of file PitchCorrector.h.

◆ getRootPitchClass()

template<FloatType T>

int dspark::PitchCorrector< T >::getRootPitchClass ( ) const

inlinenoexcept

Returns: The root pitch class in effect (0 = C .. 11 = B).

Definition at line 384 of file PitchCorrector.h.

◆ getScaleMask()

template<FloatType T>

std::uint16_t dspark::PitchCorrector< T >::getScaleMask ( ) const

inlinenoexcept

Returns: The scale mask as passed to setScale() (root-relative).

Definition at line 377 of file PitchCorrector.h.

◆ getState()

template<FloatType T>

std::vector< uint8_t > dspark::PitchCorrector< T >::getState ( ) const

inline

Serializes the parameter state (setup/UI threads; allocates).

Definition at line 403 of file PitchCorrector.h.

◆ prepare()

template<FloatType T>

void dspark::PitchCorrector< T >::prepare	(	const AudioSpec &	spec,
		int	fftSize = `0`
	)

inline

Allocates the detector and shifter state (setup thread).

Invalid specifications and non-finite sample rates are ignored and the previous state is kept; an unprepared instance passes audio through untouched. The detector uses its own automatic window policy, which holds its analysis span constant across rates; the shifter's frame follows the policy documented above.

Parameters

spec Audio environment specification.

fftSize Analysis frame for the internal shifter. Values <= 0 (the default) select the AUTOMATIC frame: the smallest power of two in [512, 32768] spanning at least 2048/48000 s (42.7 ms) at this rate - 2048 at 44.1/48 kHz, 4096 at 88.2/96 kHz, 8192 at 176.4/192 kHz. Read it back with getFrameSize(). An explicit request is rounded up to a power of two and clamped to [256, 1 << 20]; going below the automatic value buys settling time and costs the low register, in the proportions the frame note above states.

Definition at line 284 of file PitchCorrector.h.

◆ processBlock()

template<FloatType T>

void dspark::PitchCorrector< T >::processBlock ( AudioBufferView< T > buffer )

inlinenoexcept

Processes audio in-place. Pass-through until prepare() succeeds.

Channels beyond the prepared count are left untouched, as in the shifter. Detection reads channel 0; every processed channel receives the same correction.

Parameters

buffer Audio block.

Definition at line 453 of file PitchCorrector.h.

◆ reset()

template<FloatType T>

void dspark::PitchCorrector< T >::reset ( )

inlinenoexcept

Clears all signal state and keeps the parameters (stream owner).

The applied correction restarts at zero: a fresh stream has no note history worth continuing to correct.

Definition at line 314 of file PitchCorrector.h.

◆ setFormantPreserve()

template<FloatType T>

void dspark::PitchCorrector< T >::setFormantPreserve ( bool on )

inlinenoexcept

Enables the shifter's cepstral formant preservation (the anti-chipmunk envelope pre-warp). Default off.

Definition at line 371 of file PitchCorrector.h.

◆ setRetuneSpeedMs()

template<FloatType T>

void dspark::PitchCorrector< T >::setRetuneSpeedMs ( T ms )

inlinenoexcept

Sets the retune speed: the time constant, in milliseconds, of the glide from the sung pitch to the target note.

0 selects the hard snap (the full correction at once, quantized character). Values are clamped to [0, 10000]; non-finite values are ignored.

Definition at line 362 of file PitchCorrector.h.

◆ setScale()

template<FloatType T>

void dspark::PitchCorrector< T >::setScale	(	std::uint16_t	scaleBitmask,
		int	rootPitchClass
	)

inlinenoexcept

Selects the scale the output snaps to.

Parameters

scaleBitmask	12-bit harmony::NoteSet: bit k is the scale degree k semitones above the root (the harmony::allScales layout). Bits above the low 12 are ignored, and an empty mask disables correction.
rootPitchClass	Root as semitones above C; any integer folds into 0..11.

Definition at line 341 of file PitchCorrector.h.

◆ setState()

template<FloatType T>

bool dspark::PitchCorrector< T >::setState	(	const uint8_t *	data,
		size_t	size
	)

inline

Restores parameters from a blob (tolerant; rejects foreign ids). A missing field restores its default: chromatic from C, hard snap, formants off.

Definition at line 416 of file PitchCorrector.h.

The documentation for this class was generated from the following file:

PitchCorrector.h

Public Member Functions

Detailed Description

Member Function Documentation

◆ getFormantPreserve()

◆ getFrameSize()

◆ getLatency()

◆ getRetuneSpeedMs()

◆ getRootPitchClass()

◆ getScaleMask()

◆ getState()

◆ prepare()

◆ processBlock()

◆ reset()

◆ setFormantPreserve()

◆ setRetuneSpeedMs()

◆ setScale()

◆ setState()